Now, I’m not talking about bringing back Bob. In fact, Bob was the worst approach to these ideas. He embodied a person visually and then acted like the least courteous, most annoying person possible. But this doesn’t just apply to anthropomorphized interfaces with animations or video agents. All applications and interfaces exhibit the characteristics that Nass and Reeves have studied. Even before Microsoft Word had Clippy—or whatever that little pest is called—it was a problem. Word acts like one of those haughty salesclerks in a pricey boutique. It knows better than you. You specify 10-point Helvetica but it gives you 12-point Times at every opportunity. It constantly and consistently guesses wrong on almost every thing. Want to delete that line? It takes hitting the delete key three times if the line above it starts with a number, because of course it must, must be a numbered list you wanted. You were just too stupid to know how to do it. Interfaces like that of Word might be capable in some circumstances, but they are a terrible experience because they go against human values of courtesy, understanding and helpfulness, not to mention grace and subtlety.
So when you’re developing a tool, an interface, an application or modifying the operating system itself, my advice throughout development and user testing is to ask yourself what type of person is your interface most like? Is it helpful or boorish? Is it nice or impatient? Is it dumb or does it make reasonable assumptions? Is it something you would want to spend a lot of time with? Because, guess what, you are spending a lot of time with it, and so will your users.
I don’t expect devices to out-think me, think for me, or protect me any more than I expect people to in my day-to-day life. But I do expect them to learn simple things about my preferences from my behavior, just like I expect people to in the real world.
Nathan Shedroff, "Computer Human Values," Boxes and Arrows, 23 Jun 2004
